Optimizing Combination Therapy to Accelerate Clinical Cure of Tuberculosis

Drusano, George; Louie, Arnold

Abstract

Infection with Mycobacterium tuberculosis (MTB) is a massive worldwide problem. The advent of Multiply Drug- Resistant and eXtensively Drug-Resistant strains (MDR and XDR MTB) has exacerbated the problem and has resulted in increased mortality and substantial morbidity. Other than bedaquiline, which was approved several years ago but with a black box warning, the last new MTB agent was rifampin. However, lately a number of new agents, some with unique mechanisms of action have entered the developmental pipeline. This will ultimately help with the therapy of MDR/XDR MTB. A large part of the difficulty in treating MTB is the duration of therapy. Fully susceptible strains requir 6 months of therapy while MDR/XDR strains require 18-24 months of therapy or longer. Such long therapeutic durations exacerbate problems with adherence, which is a major driver of resistance. Further, particularly with MDR/XDR MTB, therapy has many second line agents which are more toxic than first line drugs. It would pay massive public health dividends to be able to shorten therapy. While we have new agents entering the therapeutic armamentarium, little thought has been given to how to use them to improve cell kill, suppress resistance and, hence, have the possibility of shortening therapy. It is the overall goal of this Program to identify optimal regimens that fulfill the requirements of shortening therapy: most rapid cell kill, resistance suppression and activity against different metabolic states in which MTB exists (log-phase growth, acid-phase growth and Non-Replicative Persistent Phenotype-phase). There are three Projects and three Cores. The Projects involve evaluating combinations of MTB drugs in the Hollow Fiber Infection Model (HFIM), in murine models of infection and in the Cynomolgus macaque Non- Human Primate model (NHP). The Cores are the Administrative Core, Drug Assay Core and Mathematical Modeling Core. All Projects and Cores will interact and cross support. The HFIM has the flexibility to study all the metabolic states and to do so with human, murine and NHP drug profiles. A publication from our lab noted that animal drug profiles alter the activity of drugs on the pathogens being modeled. There has been speculation regarding the utility of animal system for reliability to design human trials. We will use the HFIM to generate data on combination therapy kill rates and resistance suppression in each metabolic state, using human and animal profiles. The mathematical modeling will allow direct identification of the impact of the different profiles on endpoints. These HFIM estimates can then be compared to the modeled data in the animal systems. Driving effect parameters with different profiles will allow further insight into what information can be reliably extracted to allow the best bridging to human infection. Sequencing of regimens, with the follow-on regimen being targeted at the organism states remaining after the first regimen and with resistance mechanisms of the two regimens being independent may be the best way to shorten therapy. This can be a general paradigm for future combination regimen development.

Public Health Relevance

This Program Project will delineate a pathway to evaluate combinations of agents for therapy of Mycobacterium tuberculosis. This pathway will identify regimens which will kill rapidly, suppress resistance and allow shortening of duration of therapy. Project-001 - Hollow Fiber Infection Model to Delineate Combination Drug Interaction for Different MTB Metabolic Populations and using Different Pharmacokinetic Profiles Project Leader (PL): George Drusano DESCRIPTION (as provided by applicant): In Project #1, our Goal is to employ the flexible and powerful Hollow Fiber Infection Model (HFIM) to help identify optimal combination chemotherapy regimens that will provide maximal rates of bacterial cell kill and prevent resistance emergence. In so doing, we hope to markedly foreshorten the duration of chemotherapy for patients infected with Mycobacterium tuberculosis (MTB). Part of the power of the HFIM is its ability to study MTB in Log-Phase growth as well as in Acid-Phase and Non-Replicative Persister- (NRP) Phase. In this Project, all of these phases will be examined. There are a large number of possible two drug combinations. Indeed, there are too many combinations to be evaluated in the time frame of this proposal. Consequently, we will employ the fully parametric Greco drug interaction model as a method to rank order the priority with which combinations will be tested. As with the HFIM, we will test all metabolic populations in these checkerboard evaluations. The metric for ranking will consist of evaluation of the ? (drug interaction parameter) and its estimated 95% confidence interval. Larger ? values and narrower confidence intervals will be given greater weight. We will also look at the actual observed depth of cell kill and the effect parameter from the Greco model. Each of the metabolic populations will contribute 1 set of these parameter values, confidence intervals, etc. It is straightforward to calculate a metric for determining the overall rank order of combinations to be evaluated in the HFIM. We will first use human pharmacokinetic (PK) drug profiles to evaluate the drug interaction for effect (synergy, additivity [Loewe Additivity], antagonism) for each metabolic population. Here, in contradistinction to the screening assay in plates, we will also determine amplification or suppression of less-susceptible subpopulations for the agents in the combination. The PK profile has been demonstrated to have a major impact on cell kill and resistance amplification/ suppression. The use of animal systems may possibly give a misleading conclusion. We will also test these combinations in the HFIM for all metabolic populations using both murine and cynomolgus macaque PK profiles. These findings will allow direct comparison to findings in Projects #2 & #3, where these combinations will be examined in these systems, respectively. Use of mathematical simulation from the animal and from the HFIM outcomes will identify the most reliable information to be gleaned from each model. Finally, we will test what we feel to be the approach which will yield the highest likelihood of achieving major shortening in MTB therapy duration: the evaluation of two regimens that are independent by resistance mechanisms where there is a transition after maximal response has been obtained by the first set of drugs. This Project is informed by all Cores and Projects and informs all other Projects, making this a centerpiece of this highly interactive and synergistic Program Project Grant.